Predicting time to nursing home care and deathPatients with Alzheimer's disease and their families often ask how long it will be until the patient will require nursing home care or die. We have developed a technique for providing an answer to this question. This website has been designed to assist you in applying our technique to make these predictions about a patient.



Go to form for calculating predictions


For whom can you make predictions?

The closer the patient resembles those that we studied to develop our technique, the better. Our predictions are based on a study of patients with relatively mild probable Alzheimer's disease. None had a life-threatening illness when we began to study them. None had a history of stroke or signs of a stroke when a neurologist examined them.

What do I need to do to get a prediction?

Our prediction technique makes use of clinical features that a doctor can observe when examining a patient with Alzheimer's disease. The specific features we use are listed and explained below. For most features, you simply have to know whether they are present or absent. You also need to know the patients' score on a mental status examination, a brief test of memory and other cognitive functions that is typically administered by the physician when evaluating a patient with Alzheimer's disease. Once you have the required information, you can enter it on an information form. The program will evaluate the patient's information and supply the predictions. If are not the patient's doctor and you are unsure about a specific piece of information, ask the patient's doctor.

What kind of prediction do I get?

The program makes two predictions. The first is the number of months until the person requires nursing home care, that is the level of care provided by a nursing home. The second prediction is the number of months until the patient dies. These are not exact predictions. They take the following form:

"Based upon the patients that we have followed, 25% of patients with a similar clinical profile require nursing home care within __ months, 50% within __ months, and 75% within __ months."


"Based upon the patients that we have followed, 25% of patients with a similar clinical profile die within __ months, 50% within __ months, and 75% within __ months."

Note that we are not giving an exact prediction. Rather, we are providing a time range for the outcome.

How accurate are these predictions?

This is the first time that a technique to make these predictions for individual patients has been developed. We did try out the approach on it to a different population from the one it was developed with, and it worked quite well. Still, we see it only as a first, promising step and hope to do more work to increase our confidence in its accuracy. At the least, this approach can differentiate patients with better and worse prognoses to a degree that is not presently possible.

Where can I read more about how you developed this technique?

This technique and its derivation are described more fully in an article that we published in the Journal of the American Medical Association (Stern Y, Tang MX, Albert M, Brandt J, Jacobs DM, Bell K, Marder K, Sano M, DP Devanand, Albert S, Bylsma F, Tsai WY. Predicting time to nursing home admission and death in individuals with Alzheimer's disease. JAMA 1997;277:806-812.)

Go to form for calculating predictions

This work is a product of the Predictors Study. This study, now in its eighth year, is specifically designed to investigate the natural history of Alzheimer's disease, in order to develop predictor models for use in the management of the disease. To that end we have been following over 230 patients with Alzheimer's disease every 6 months since the start of the study. The study is directed by Yaakov Stern, PhD at Vagelos College of Physicians and Surgeons in New York. Investigators at two other institutions collaborate: Marilyn Albert, PhD at Massachusetts General Hospital and Harvard Medical School in Boston, and Jason Brandt, PhD at Johns Hopkins University School of Medicine in Baltimore.

This study is supported by the National Institute of Aging (RO1-AG07370). It is housed in the G.H. Sergievsky Center and the Taub Institute for Research on Alzheimer's and the Aging Brain at the Columbia University Irving Medical Center, New York, NY.

For comments, suggestions or questions, write to:
Yaakov Stern, PhD
Columbia University College of Physicians & Surgeons
630 West 168 Street
New York, NY 10032

Last December 20, 2019